It’s interesting to read what everyone is seeing from their thoughtful vantage points, and thank you for posting. I appear to look at different kinds of volume data than others, so I’ll throw in my troubling inflation-adjusted two cents.
The volume of share sales is moving downwards in the manner of steps on a fairly precisely measured incremental staircase, accompanied by violent upswings in prices over several minutes (in the Index futures markets). No desperate regurgitation of lunch by the institutions at this point. If anything, the selling is more methodical than it has been in recent days. As mentioned in what I posted about the shock of unanticipated inflation results several weeks ago, traders should be careful about the valuations of shares selected, because the markets are especially unforgiving on high-multiple businesses in these environments. In the 1970s, entire columns of high-quality names were trading at single-digit multiples once all the first-class passengers’ lunches were redistributed.
The Simple Bottom Indicator (unsurprisingly) flashed a ‘Minor Bottom’ also. It was a mid-strength minor signal, circa 2/3rd’s the way towards a major signal.
In terms of price action 2022 appears more like a ‘regular’ bear market (vs. say 1987 or 2020) where (as posted above) the expectation (based on past behavior) is that there’ll be a number of steps downward until there’s a large ‘panic drop’ where I expect there are multiple days of ‘Major Bottoms’. Once the market stabilizes after that phase is normally an OK time to invest. The market may proceed lower from there but typically not too far.
One observation - that may play out or not as this develops - is that this ‘bear market’ appears to be playing out at a faster tempo than say 2007-2009. The drops & bounces each appear to happen in a shorter time-frame. Nothing concrete - I’ve not bothered to calculate the difference etc. - just an interesting curiosity from chart eye-balling.
FWIW my bottom detectors are totally silent. Major bottom, short term bottom. Nada.
They have as one requirement that things be bad for at least a few days on one or two metrics,
on the theory and historical observation that it takes a while for true capitulation to set in.
You can have quite nice market bottoms without those measures of capitulation, but that’s what my models look for.
Do you have examples of major bottoms without capitulation?
Well, using my model’s definition of capitulation:
It would be the times my models failed to signal a bottom.
But I have tuned them to be less and less discriminating till all bottoms are found, in return for more false signals on the way down.
I’ll have to turn the sensitivity knob to reduce the number of false signals then see which major bottoms get missed.
…[pause]…
These are the first few bottoms to be missed if I increase the discrimination knob a bit:
1978-03-01
1982-08-09
and both dates of the double bottom 2002-10-09/2003-03-12.
Obviously these were points of capitulation—just not in a way that my model detected very well.
For a sense of market conditions on those days:
New lows were only 12.1% of NYSE issues on 1982-08-09, arguably the best moment in history to be a buyer.
Yesterday’s figure was 23.2%
They have as one requirement that things be bad for at least a few days on one or two metrics, on the theory and historical observation that it takes a while for true capitulation to set in.
As mentioned above, I think what I’ve come up with would be more accurately referred to as a ‘bounce detector’. And even then, it sometimes takes several signals over a period of a few days to result in a bounce.
Mungo, I know your bottom detectors are far better tested and more meaningful and I appreciate you sharing your signals!
Mungo, I know your bottom detectors are far better tested and more meaningful and I appreciate you sharing your signals!
Zeelotes’ are much better than mine.
But I try : )
I haven’t really changed them in a while. I’ve added a couple of slight variants, but never removed the original.
Which means they aren’t any better than they were a decade ago, but on the other hand they have some real world out-of-sample experience.
Much easier than designing a top detector, that’s for sure.
For a buy signal:
Use this chart https://stockcharts.com/h-sc/ui?s=%24BPNYA&p=D&yr=1&…
Wait for the level to get below 31 (it has, as of today).
Then wait for it to rise by 6% from its low (meaning 6% of the value at the low, not 6 ticks) some time in the last 5-6 trading days.
Shortly after a market bottom, especially Nasdaq-related, you’ll generally get a few days of buy signals in a row.
In the current dip, so far the lowest value is today’s at about 28.76 and counting.
A rise of 6% would give a buy signal at a rise of above about 30.5.
But if/as the current level hits new lows, the “6% rise” target will also be lower.
How are you supposed to invest in short term bottom? Especially with bear catchers at zero. Is there a short term top finder? Or just stay in for a month?
How are you supposed to invest in short term bottom? Especially with bear catchers at zero. Is there a short term top finder? Or just stay in for a month?
It depends on the signal–both its usual useful time frame, and its average reliability.
Statistically, my major bottom detector gives excellent times to start long term positions.
Sometimes it’s too early so the first couple of months are unpleasant as the market drops further, but the one year forward returns tend to be good.
That’s the type I find most useful: it’s the date to blow the cash you were saving for just the right time to buy good stuff on sale.
Other bottom detectors are good for time frames of a week or a month, the short term bottom detectors you mention.
They also tend to be a good time to deploy cash you have on the sidelines.
But you would want to keep an eye on things—if you don’t get some other bullish signal
(confirmation) within a couple/few weeks, you might want to back out again after a quick profit.
They can also be good for things that you wanted to invest in for a while, but were just waiting for a good price on an auspicious date.
For example, that’s how I used the most recent one: I bought some Alphabet stock that I had been planning to buy as soon as it looked like a good moment to pounce.
I plan to keep the position for a while, but at least a short term bottom signal gives me an idea that I bought on a better-than-average day to be a buyer.
Yet others are good for “as long as they stay bullish”.
You don’t know if that’s a couple of weeks or a year or more, you just keep checking.
I don’t know of a good short term top finder, other than various trend following indicators rolling over and starting down a gain.
Something based on breadth is probably best, since failed rallies tend to narrow rapidly.
The “acid test” is something that found the short term top in January 2009.
People forget the magnitude of the rally in late 2008, and the magnitude of the secondary plunge to the lower March lows.
Wait for the level to get below 31 (it has, as of today). Then wait for it to rise by 6% from its low (meaning 6% of the value at the low, not 6 ticks) some time in the last 5-6 trading days. Shortly after a market bottom, especially Nasdaq-related, you’ll generally get a few days of buy signals in a row.
There was a low(er) low on May 12th at 24.99, and it moved up above the 6% threshold (26.49) the next day. SPY did relatively well for almost one month, but fell down below a May 13th MOO trade on June 10th.
Is this one of those times where “signals are wrong sometimes”?
<People forget the magnitude of the rally in late 2008, and the magnitude of the secondary plunge to the lower March lows.> … March 09 was truly terrifying.
It was. I could claim calm, but I’d be fibbing.
I was a paid money manager at that time and gave good advice to my clients, but it didn’t come from a place that was calm, cool, and collected.
It was particularly terrifying to those who figured the December/January rally was the all clear and removed all defensive positions of whatever type.
I think of it as “the forgotten bull market”.
The Russell 1000 (IWB) was up over 27% from its low 2008-11-20 till the temporary peak 2009-01-05.
There was a low(er) low on May 12th at 24.99, and it moved up above the 6% threshold (26.49) the next day. SPY did relatively well for almost one month, but fell down below a May 13th MOO trade on June 10th.
Is this one of those times where “signals are wrong sometimes”?
Wrong might be harsh. Maybe just “not very long lasting”.
I don’t know this signal well.
The analysis Zee did presents the information the way he looks at it:
What is the average gain/loss for various time frames after each signal?
It’s pretty strong–on average–at quite a few time frames.
It’s quite possible that the recent signal was a good one, and this week is just a transient dip in the rally it portended, creating a second signal.
Not my first guess, but we’ll know soon enough.
I think it’s more likely that it’s a signal you want to believe only as long as some indicator is still bullish.
Maybe the same signal, maybe a separate sanity check signal.
So maybe the buy is a state, not a date.
For a buy signal: Use this chart https://stockcharts.com/h-sc/ui?s=%24BPNYA&p=D&yr=1&…… Wait for the level to get below 31 (it has, as of today). Then wait for it to rise by 6% from its low (meaning 6% of the value at the low, not 6 ticks) some time in the last 5-6 trading days. Shortly after a market bottom, especially Nasdaq-related, you’ll generally get a few days of buy signals in a row.
This woulda worked pretty darned well in March 2020. If you didn’t quite beleive it on March 17th, you got another chance to climb aboard on March 24th.
Jim wrote: Speaking of bottom detectors, don’t forget this one from Zeelotes from (yikes) 14.5 years ago.
Here is an update using 31 as the threshold and requiring a 6 percent rise. These are the forward returns in the Nasdaq 100 from each signal date with repeats within six market days removed.
Here is an update using 31 as the threshold and requiring a 6 percent rise. …
Great data, thanks.
Just to make sure I understand the table, are these correct interpretations:
Max Pop 3 is the largest rise from signal date to the highest level touched within 3 months of the signal date
Max Drop 3 is the largest fall from the signal date to the lowest level touched within 3 months after the signal date.
Fixed Fwd 2 is simply the return in the exact 3 months starting on the signal date
First row is the average of figures in the column below